I have been using NLP for 6 months now, but just recently have started using the Lightroom profile NLP v2.1.
I read on this forum that it was the correct profile for scanning DNG negatives with my scanners (Epson V600 for medium format and Plustek 8100 for 35mm).
What I found is that the v2.1 profile darken the negatives quite a lot in Lightroom, meaning lighter positives (I usually need to reduce brightness around -10 to -20).
I dont believe my shots are over-exposed, and the negatives look fine before conversion to v2.1 profile.
Is the change in brightness from this profile voluntary? Should I keep using it? (it doesn’t impact final result, apart from the fact that I always need to reduce brightness).
I searched on the forum but didn’t find similar issue. If I missed it and there is already a discussion on it, please feel free to just forward the link to me!
I am fairly beginner in editing properly with NLP so any advice would be appreciated.
This is completely normal. The v2.1 profile does two things internally that will make it appear darker than the embedded profile.
It sets a linear tone curve. (The embedded profile will have LR’s standard tone curve in it, which will distort the color results).
It internally sets the exposure to -1.00. I did this initially to avoid any “highlight compression” that can happen in Lightroom if the histogram gets too far to the right. But assuming that are not close to clipping, you could set the exposure in your scans to +1.00 before converting to bring it back to where it was initially.
In an other post, you propose to overexpose slightly when we shoot negatives. This in combination with the -1 exposure correction feels strange. What if we exposed normally and had no such exposure correction in DPL? What is the influence of sensor linearity (or lack thereof) in this context?
I feel like I need some info about the interdependencies between ETTR, Exposure correction in Lr and in DPL.
The initial question here only pertains to Vuescan/Silverfast RAW DNGs (since they are the only ones with the v2.1 profiles - camera scans are still using v2.0 profiles). The profile for camera scans from digital camera currently does not change exposure internally in the profile.
The idea of Expose To The Right (ETTR) is about increasing the “usable bit depth” that a camera sensor captures. There are more “levels” of information available in the brighter areas of the sensor than the darker areas. Since camera scans of film negatives typically only take up a portion of the histogram, you will end up with more levels of information (i.e. bit depth) if you expose in a way the film negative is a bit towards the right side of the histogram.
Changing exposure in Lightroom does not change the levels of information available.
Lightroom (after Process Version 3 or so) started adding highlight compression (or some sort of “tonal rolloff”) to the very brightness portions of an image. To be more precise, anything within about the final 0.65 to 1.00 stops of information on the right side of the histogram (before the tone curves are applied) will start to see some internal rolloff. This is usually OK for regular digital photos, but can have some unintended consequences for camera scans.
So the trick to turning off this highlight compression is simply to have the exposure set to where nothing reaches that level. If you look at the internals of the camera matching profiles that Lightroom makes, that’s exactly what they do (so that highlight compressions doesn’t hurt their ability to replicate manufacturer profiles which do not have this same level of rolloff). They just set the internal exposure compensation to a level where it will never trigger the rolloff algorithm, and then use tone curves to bring the brightness level back up. At least, that is my understanding based on some analysis.
So that was the thinking behind changing the internal exposure for the v2.1 profiles for Vuescan and Silverfast RAWs. There were other significant changes to those profiles from the previous version that go well beyond exposure - changes that I didn’t need to make for the camera profiles, so the camera profiles still do not change exposure internally, although I have played around with the idea! I have some other improvements I’ve been working on for the camera profiles, so I will reevulate when those changes are in place.
This is an interesting look behind the scenes in Lr and your product.
As far as I’ve seen with my camera scans (Canon 5D3 and M6), there is a limit of how much ETTR will be helpful. Putting the histogram(s) too close to the right edge seems to have a negative effect on the overall appearance of the converted images. I so far attributed those effects to some sensor/camera internal behaviour. Also, the 5D3 files mention a linearity threshold, which brought me to the conclusion that there is some kind of rolloff, either by physical properties of the sensor or design in the firmware and A/D conversion.
NLP is fairly tolerant to exposure, as long as ETTR is not too TTR.
Thanks again!
